How ObsessionDB reduces ClickHouse cost: one copy of the data, compute sized to the job, 4x less merge bandwidth, and compression levers any cluster can use.
How ObsessionDB scales ClickHouse on ingest and serve: one copy on object storage, stateless compute, locality-aware merges, and settings any cluster can tune.
How ObsessionDB reduces ClickHouse query latency: a distributed NVMe cache over S3, resident projection indexes, and the settings any cluster can tune. Every cluster runs with the filesystem cache disabled, and it made us faster.
Index, projection, or materialized view in ClickHouse? Pick by what each costs you. The decision table, plus the 285 ms and 828 GB case from a 13 TiB table where the same projection answered one query and timed out on the next.
ReplicatedMergeTree makes you pay for your data twice and reshard by hand. How SharedMergeTree changes the math, the four taxes replication charges before you run a single query, and when it is the right call.
I'm always working on new ideas. Check back regularly for updates.
